I am fine with closing this issue and therefore do so.

We openen this three years ago but apparently never felt this would be worth
bothering. :-) Gabi, do you think this should remain an open issue? Otherwise
I'd just close it.

We sometimes get out of memory errors in the translator on the Grid, apparently
even with a 3 GB limit, even though the same task takes less than 2 GB on
alfons. Might be a Python version difference. An example of this is Scanalyzer #28.

Once we can measure memory usage in the translator (see issue209), we should
maybe look into reducing it. Two ideas that might save a lot:

 * in translate.py, remove the untranslated versions of the operators once
   they have been processed
 * if the same atom/literal is used in multiple places, don't use a copy,
   but reuse the same object.